MAANG companies (excluding N) are requiring relocation and RTO (hybrid, near an office). Also, their tone has shifted to where their recruiters feel like they hold a monopoly on in the business relationship power dynamic.

Perhaps shrinking corporate real estate holdings of properties and lease should be prioritized over burdening employees with additional relocation, commuting, and housing costs.

I don't think most people appreciate how far away it is to reach Mt. Hamilton or how precarious the road up it is. Took a couple of gfs on drives up there.

I used to bike up Hicks Road near Mt. Umunhum.

Other good vantage points: the top of The Dish route behind Stanford and the observation deck of the de Young Museum.

0. https://www.teamblind.com/post/Does-Netflix-do-remote-hires-...

Step 0. Don't bother with obsolete Cat 6 or lower cable because it's just not worth the effort over Wi-Fi. If you're going to do it, buy a 1000' spool of CMP Cat 8 UPoE-rated premise wire (~$1000-$1500 USD), punch down tool, J boxes, punch down receptacles, semi-flexible plenum large conduit that doesn't have ridges, cable fish tape, wall-mounted quarter 19" rack, 90 degree drill and hole saw with various lengths of extensions (usually from about 1' to 10' or longer for tall walls), and a punch down patch panel.

Step 1. Run conduit and J boxes first to a central closet or to a garage in most instances. This is the labor-intensive, hard part but it makes running multiple cables through it much easier. Larger homes may need a second closet or second conduit hub area and more conduit to a main patch panel.

Step 2. Easily slide wire in using a fish tape, 1 to 4 cables per box depending on the location. It's so much easier with semi-flexible conduit.

Step 3. Test each cable using iperf3 with an RPi and laptop, or borrow a commercial 10GbE cable tester like a DSX2-8000.

Step 4. Drop in a 10 GbE U/POE++ switch that doesn't sound like a jet engine.

Option: Instead of, or in addition to, Cat 8 wiring, run fiber.

Around 1994, I managed to get to a Computer Literacy bookstore.

With a Windows Enterprise license, all of the ads can be turned off, or at least that was the practice.

Here's a naive, partial implementation that runs in <500 ms so long as there's an existing dnf cache:

   command_not_found_handle() {
    local pkgs
    readarray pkgs < <(dnf rq --whatprovides "$1" -C --qf '%{name}\n' 2>/dev/null | sed '/^$/d' | sort -uVr)
    if (( ! "${#pkgs[@]}" )); then
      echo >&2 'Command not found and no package provides it according to the dnf cache'
      return
    fi
    echo >&2 'Command not found, but offered via the following command(s):'
    pkgs=("${pkgs[@]/#/    dnf install }")
    echo >&2 " ${pkgs[@]}"
  }

This trend has nothing to do with productivity and is all about the egos of control-freak managers "maintaining control" over employees in a way that wastes employees' time and money and harms the company's reputation.

The problem you describe is solved by introducing capabilities such as in seL4. Without beginning with a capability, something cannot do something else.

The solution is there need to be parts available for reasonable prices in the first place. There are no reasonably priced parts for iPhone Pro 13 and onwards. My friend's cell phone repair chain has lost a lot of business because they can't fix many types of Google Pixel, Samsung, or Apple devices for reasonable prices.

Can't someone !w piratebay and click the link?

And when running a shell, do it like this:

    docker run user/image:tag -it --rm bash -li

My mantra is all programs and widely-used company tools should have current man pages and shell completions. The whole "Go look at a web page" for "help" is slow, distracting, and lazy.

Sounds like something Numba should include perhaps.

I retired from HPC around the era of iPython, Open MPI, Infiniband, OFED, GPFS/Panasas/Lustre, Rocks Cluster, and Condor/PBS/SGE.

On the first part, if a skill doesn't support a precise sentence structure, it becomes difficult to use or simply doesn't support features users expect to use using natural language.

On the second point, not all Alexa devices work with IPv6. I have an Echo Show 15 here that simply refuses to play BBC News and other "Flash Briefing" items when IPv6 is enabled. Really? It's 2024 now. Did the whole factory reset, reboot, and so forth dances.

An LLM startup with enough partner biz dev/sales talent could combine IoT API integrations from enough manufacturers to side-step the boring and limited Apple Siri, Google Hello, and Amazon Alexa oligopoly hegemonies and self-destructive platforms like IFTTT to execute on a compelling, competing replacement making good on all 4 points. However, I don't think giving away computing resources for free is necessarily a scalable business model but it's fair to charge a reasonable micropayment subscription to host and run your code and data.